DART: Nudging an Asteroid by Hitting It

580 kilograms at 6.1 kilometres per second

DART launched on a Falcon 9 from Vandenberg on 24 November 2021 and cruised for ten months to Didymos, an Apollo-class near-Earth binary found in 1996 by the Spacewatch survey. Didymos is a top-shaped body about 780 m across, spinning once every 2.26 hours, with a mass near 5.3 × 10¹¹ kg. Its moonlet Dimorphos — inferred from lightcurve dips in 2003 by Petr Pravec and colleagues, named in 2020 — is an oblate lump roughly 151 m across, circling its primary every 11.92 hours at about 1.19 km.

The spacecraft carried one instrument, DRACO (Didymos Reconnaissance and Asteroid Camera for Optical navigation), feeding an autopilot called SMART Nav. Dimorphos resolved as a separate dot only about 50 minutes before impact, and the autopilot re-targeted onto it with no human in the loop — round-trip light time to Earth was some 75 seconds. The last complete frame arrived about two seconds before contact from roughly 12 km out, at a few centimetres per pixel, showing a surface carpeted in boulders.

Impact occurred at 23:14 UTC on 26 September 2022, about 11 million km from Earth, with the reconstructed spacecraft mass at impact 579.4 kg and the closing speed 6.14 km/s. The kinetic energy was about 1.1 × 10¹⁰ J — the equivalent of roughly 2.6 tonnes of TNT. That is a firecracker on planetary scales. What mattered was not the energy but the momentum, and above all what the impact threw away.

Momentum enhancement: why β is bigger than one

If a spacecraft of mass m hits an asteroid at speed U and simply embeds itself, the target gains momentum mU and nothing more. Real impacts excavate a crater, and the excavated rock leaves the surface moving outward — roughly back along the direction the impactor came from. Every escaping kilogram carries momentum away, and by Newton's third law the asteroid recoils. The delivered momentum is therefore

Δp = β m U,   with   β = 1 + p_ejecta / mU

β = 1 means every scrap of debris falls back and the impactor is just a bullet. β > 1 means the crater acted as a rocket nozzle. For Dimorphos, Andrew Cheng and the DART investigation team derived β = 3.61 (+0.19 / −0.25) — the ejecta carried roughly two and a half times more useful momentum than the spacecraft did.

Nothing is created here. Momentum is exactly conserved: plume plus asteroid equals what the spacecraft brought. β counts only the asteroid's share along the impact direction, which the recoil inflates. Energy is emphatically not conserved — most of the 11 GJ went into crushing, heating and fragmenting rock.

What sets β is the target, not the impactor. Two properties made Dimorphos an extravagant performer:

• It is a rubble pile. A loose, low-cohesion heap of boulders and grit launches easily instead of resisting as a monolith. Highly compressible, very porous targets do the opposite — they compact under the shock, swallow the energy and retain their ejecta, pushing β back toward 1.
• Its escape velocity is absurdly small. With an estimated mass near 4.3 × 10⁹ kg and a radius of ~75 m, escape speed is about 9 cm/s. On Earth, crater ejecta rains back down; on Dimorphos, anything faster than a strolling ant is gone forever.

Impact-scaling theory gives the ejecta mass moving faster than v as roughly M(>v) ∝ v^−3μ, with μ ≈ 0.4 for porous material and ≈ 0.55 for competent rock. The exponent is steep, so the momentum integral is dominated by the slowest, most massive ejecta — exactly the material a 9 cm/s escape velocity lets go. Order 10⁶ kg of rock at metres per second explains the observed boost. One subtlety: the ejecta cone leaves along the local surface normal, not the impactor's track, so the extra push is a vector not parallel to the incoming spacecraft.

From 2.7 mm/s to 33 minutes: the orbital arithmetic

Dimorphos orbits Didymos at a semi-major axis of about 1189 m with a period of 11.9216 h (42,918 s). Its orbital speed is therefore 2πa/P ≈ 17.4 cm/s — slower than a brisk walk. DART hit almost head-on into that motion, so the delivered impulse subtracted from the orbital velocity. Cheng et al. put the along-track change at Δv = 2.70 ± 0.10 mm/s, consistent with β m U ≈ 1.3 × 10⁷ kg m/s spread over the moonlet's mass.

For a near-circular orbit, a small tangential impulse changes the semi-major axis by Δa/a = 2Δv/v, and Kepler's third law (P² ∝ a³) then gives ΔP/P = 1.5 Δa/a. Combining:

ΔP / P = 3 Δv / v

Plugging in: 3 × (0.0027 / 0.174) = 4.7%, and 4.7% of 715 minutes is about 33 minutes. The semi-major axis shrank by roughly 37 m. The measured value — a period change of −33.0 ± 1.0 minutes, from 11 h 55 min to 11 h 22 min — matches that back-of-envelope calculation almost exactly. Note the sign: a slower moon falls to a lower orbit and therefore goes around faster. Slowing something down speeds up its year.

The declared minimum success criterion was 73 seconds, chosen as comfortably detectable from the ground within weeks; DART beat it by about 27×. In the following weeks the period drifted a further ~34 seconds before settling — plausibly slow ejecta and boulders raining back down — and the orbit picked up a small eccentricity, with the moonlet's rotation likely knocked into an excited libration or tumbling state.

How the deflection was actually measured

At 11 million km the pair's 1.19 km separation subtends about 0.02 arcseconds — unresolvable from Earth. Astronomers never saw two dots. They measured the period from mutual events: each orbit, Dimorphos passes in front of Didymos or slips into its shadow, and the combined brightness of the unresolved point dips. Time the dips before and after, and the period change falls out.

• Ground-based photometry. A worldwide campaign — the 1 m Swope at Las Campanas, the 2.4 m Magdalena Ridge telescope, the Danish 1.54 m at La Silla, the Lowell Discovery Telescope — recovered the new period within days. NASA announced a preliminary 32-minute result on 11 October 2022.
• Planetary radar. The 70 m Goldstone antenna (DSS-14) transmitted and the 100 m Green Bank Telescope received, producing range-Doppler images that independently confirmed Dimorphos had drawn closer to its primary — the surviving radar pair after Arecibo's collapse in December 2020.
• LICIACube. The Italian Space Agency's 14 kg cubesat, released 15 days before impact, flew past at ~55 km some three minutes after contact and photographed the ejecta cone with its LEIA and LUKE cameras — direct imaging of the very plume that carried the extra momentum.
• Hubble and JWST. Both watched the impact; Hubble then tracked a dust tail that grew past 10,000 km and split in two, and David Jewitt's group counted 37 boulders up to ~7 m across drifting off at ~0.3 m/s.

Five DART papers were published together in Nature on 1 March 2023: Daly et al. on the impact site, Cheng et al. on β, Thomas et al. on the period change, Li et al. on the ejecta tail, and Graykowski et al. on the global observing campaign.

Why a millimetre per second is enough

A push of 2.7 mm/s sounds derisory, and applied at the last minute it would be. Deflection is a game played against time. A small along-track velocity change alters the orbital period, and the resulting position error grows linearly with every subsequent orbit. To good approximation the along-track miss distance after time t is

Δs ≈ 3 Δv t

So 1 cm/s applied ten years ahead yields Δs ≈ 3 × 0.01 × 3.2 × 10⁸ ≈ 9,500 km, comfortably more than an Earth radius. The same 1 cm/s applied ten days out moves the asteroid 26 km, which is nothing. This is why serious planetary defence begins with surveys, not spacecraft — the Catalina Sky Survey, Pan-STARRS, the now-retired NEOWISE, the Vera C. Rubin Observatory and, later this decade, NEO Surveyor exist to buy lead time. It is also why gravitational keyholes matter: small regions in the encounter plane that funnel an asteroid onto a later collision, as mapped for 99942 Apophis after its 2004 discovery. Missing one may need only a centimetre per second, decades early.

What DART was not

Several persistent confusions are worth clearing.

• It did not blow anything up. The specific energy delivered was 1.1 × 10¹⁰ J spread over 4.3 × 10⁹ kg — about 2.6 J/kg. The catastrophic-disruption threshold Q*_D for a body this size is of order tens to hundreds of J/kg, so DART was one to two orders of magnitude below shattering it. Reshaping, yes; destruction, no.
• It barely touched the orbit around the Sun. Essentially all of the impulse went into the mutual orbit; spread over the pair's combined mass of ~5.3 × 10¹¹ kg, the system's centre of mass changed velocity by only ~2 × 10⁻⁵ m/s. Didymos is classified as a potentially hazardous asteroid on size-and-proximity grounds, but it is not on a collision course and the test could not have put it on one.
• β is not a universal constant. It is a property of the target: its porosity, cohesion, density, surface geometry and the impact angle. A strong monolith might return β barely above 1. Quoting 3.6 for the next asteroid would be a mistake.
• It was not Deep Impact. In July 2005 NASA drove a 370 kg copper-ballasted impactor into comet Tempel 1 at about 10.2 km/s, and JAXA's Hayabusa2 fired a 2 kg copper Small Carry-on Impactor into Ryugu in April 2019. Both were cratering and composition experiments; neither measured, or attempted, an orbit change.
• It was not the movie plan. Nuclear disruption risks turning one trackable object into a poorly characterised swarm, some of it still Earth-bound. The serious version is a nuclear standoff burst, which ablates a surface layer rather than fragmenting the body — and it remains untested.

The open questions, and what Hera will settle

The single largest uncertainty in β is not the momentum — that came straight from the measured period change — but Dimorphos's mass. It was inferred by assuming a bulk density near 2400 kg/m³, and β scales directly with whatever mass is assumed. Get the mass wrong by 30% and β moves by 30%.

That is the job of ESA's Hera, launched 7 October 2024, which took a Mars gravity assist in March 2025 and arrives at Didymos in late 2026. Led by principal investigator Patrick Michel with mission manager Ian Carnelli, it will weigh Dimorphos to a few per cent by radio tracking of its own trajectory and deploy two cubesats: Juventas, with a low-frequency radar to sound the interior, and Milani, for spectral and dust work. Its targets are the questions DART could not answer:

• Is there a crater at all? Simulations by Sabina Raducan and colleagues (2024) suggest that for a body this weak and cohesionless, DART may have globally deformed and resurfaced Dimorphos rather than leaving a tidy bowl. Hera will look.
• How does β scale? DART sampled exactly one point in a parameter space of porosity, strength, impact speed and angle. Turning one data point into a predictive law is the whole practical value of the exercise.
• What state is the moonlet in? Whether Dimorphos is tumbling, how its orbit evolves under tides and the BYORP effect, and where the boulders went — some studies suggest fragments could reach Mars within ~10⁴ years, and Earth later as a harmless meteor shower.

DART proved the geometry, the autonomous navigation and the momentum bookkeeping all work. Hera turns a successful stunt into calibrated engineering.

Ways to move an asteroid, and where DART's kinetic impactor sits among them.

Method	How the push is delivered	Typical velocity change	Maturity and lead time
Kinetic impactor (DART)	Spacecraft momentum plus the recoil of escaping crater ejecta (β ≈ 1–5)	~mm/s on a 10⁹ kg body per ~600 kg impactor	Flight-proven in 2022; needs years to decades of warning
Nuclear standoff burst	X-rays and neutrons vaporise a thin surface layer that flies off like a rocket exhaust	cm/s to tens of cm/s — roughly 10–100× a kinetic impactor per kilogram launched	Never tested in space; reserved for large bodies or short warning
Gravity tractor	A hovering spacecraft tows the asteroid with its own gravity	~10 µm/s to ~1 mm/s per year, scaling with tractor mass and hover distance	Untested; needs decades, but exquisitely controllable for fine trimming
Ion beam shepherd / laser ablation	A thruster plume or laser spot ablates the surface continuously from a distance	~mm/s over months of continuous operation	Concept stage; no flight demonstration
Disruption (fragmentation)	Deliberately break the body so the pieces spread and mostly miss	Not a velocity change — it changes the problem	Last resort for very short warning; risks a shotgun spread of fragments